CHAPTER II
HISTORICAL
Chirality of a molecule was first reported in 1815 by the French physicist JeanBaptiste Biot. He found that α-quartz rotated the plane of polarized light91. However,
the first chiral separation, which laid the foundation for stereochemistry, was reported in
1848 by Louis Pasteur92-93. Because of the hemihedral facets on the crystals of racemic
sodium ammonium tartrate, he was able to separate the mirror image crystals of the
isomers by the use of a magnifying glass and tweezers. Characterization of the physical
properties of individual enantiomers, (in which the only difference lies in the opposite
rotation of a plane polarized light) led Pasteur to postulate that the enantiomers have
different three-dimensional arrangements and on the macroscopic and microscopic levels
they are mirror-images of each other94. Furthermore, he advanced the field by studying
the influence of one chiral compound upon another and introduced the technique of
resolution via diastereoisomer formation95.
This separation of enantiomers by
diastereomer formation is the basis of many modern chromatographic separation.
In 1874, the Dutch physical chemist Jacobus Hendricus van’t Hoff96 and the French
chemist Achille Le Bel97 independently theorized that the molecular basis of chirality that
was first observed by Pasteur was an asymmetric carbon.
The asymmetric carbon
proposed by Van’t Hoff had the correct tetrahedral shape, whereas Le Bel, proposed a
square pyramid.98
It is interesting to note that Pasteur’s discovery of spontaneous enantiomeric
resolution applies only to rare cases in which each isomer crystallizes separately and in a
recognizable morphologic form. For more than a century, spontaneous resolution, as well
as diastereomeric separation and differential enzymatic reactivity were the only methods
employed for enantiomeric separations. In all of these techniques the separation and
characterization of enantiomers require a chiral environment.
9
2.1 Chromatographic Methods
There are various methods for enantiomeric analysis which do not require the
separation of enantiomers.
These include polarimetry, nuclear magnetic resonance,
isotopic dilution, calorimetry, and enzyme techniques. The disadvantages in all these
techniques, however, are the need for pure chiral samples and their relative slowness. a
typical analytical problem requires separation and quantitation of enantiomers and
sometimes identification of the levorotatory or dextrorotatory enantiomer.
Both gas
chromatography (GC) and high performance liquid chromatography (HPLC) provide fast
and accurate methods for enantiomeric separation and allow quantitation of both mass
(and optical rotation for HPLC) if appropriate detection devices are used.
Chromatographic methods are considered the most useful for chiral separation99.
There are two approaches: indirect, which utilizes derivatizing agents, and direct, which
uses chiral stationary phases or chiral mobile phase additives. Chromatographic chiral
separation dates back in 1939 when Henderson and Rule100 demonstrated the separation
of d,l-p-phenylenediiminocamphor on d-lactose.
Both Kotake101 et al. in 1951, and
Dalgliesh102 in 1952 utilized paper chromatography, a cellulose support, to separate chiral
amino acids.
Although Dalgliesh was not the first to observe such separations, he
correctly attributed the enantioselectivity to adsorption by cellulose and proposed the
Three Point Rule for asymmetric recognition. This rule was later restated by Pirkle103 as
“Chiral recognition requires a minimum of three simultaneous interactions, with at least
one of these interactions to be stereochemically dependent.”
In 1960, Klem and Reed104 first reported the use of a silica gel support for chiral
HPLC chromatographic separation. Racemic naphthyl ether and phenanthrene were
partially separated on silica gel that was impregnated with (+)-α-(2,4,5,7-tetranitro-9fluorenylideneamino-oxy)propionic acid.
The first successful GC direct enantiomeric separation was reported by Gil-Av,
Feibush, and Charles in 1966105. The first stationary phase, N-TFA-L-isoleucine lauryl
ester, resolved derivatized amino acids as N-TFA isopropy, 2-butyl, and t-butyl esters,
10
with the n-alkyl esters only partially resolved. From then on, other chiral stationary phases
for GC106 and LC107 enantiomeric separations have been synthesized and developed.
2.1.2 Indirect Chromatographic method
In the indirect method, a racemic mixture is made to react with a chiral reagent to
form a pair of diastereomers and then chromatographed using an achiral column108.
Because diastereomers possess different physiochemical properties, they can be separated
in an achiral environment. The advantages of the indirect approach are the following: (1)
less expensive, i.e., conventional chromatographic columns can be used; (2) flexible
because various achiral columns and mobile phase conditions, as in HPLC, can be used;
(3) numerous types of derivatization chemistry are available and the cost of each reagent
may be less expensive than for a chiral column; and (4) different selectivities can be
achieved. On the other hand, the disadvantages of this method are: (1) long analysis time
that include sample preparation and verification of the derivatization chemistry; (2)
inconvenience, specifically in preparative chromatography, when reversal of derivatization
is needed to recover the pure enantiomers109; (3) the need to synthesize non-commercially
available pure derivatizing reagent; and (4) biased results for enantiomeric composition110
due to partial racemization of derivatizing agent or unequal reaction rates.
The indirect method of separating enantiomers in GC or HPLC has been used for
many years and is still commonly performed for various purposes111. For example, in the
bioanalytical field, the indirect method is frequently used because separation of
diastereomers from a complex matrix on nonchiral HPLC is simplified by the greater
flexibility in selecting appropriate stationary and mobile phase conditions.
Detailed and comprehensive treatments of the chromatographic separation of
diastereomers are given by Lindner112 and Souter113.
11
2.1.3 Direct Chromatographic Methods (Chiral Mobile Phase Additives)
Direct separation of enantiomers on an achiral column using a chiral mobile phase
additive is applied only in HPLC. In GC the mobile phase is an inert carrier gas, where the
possibility of selective interactions with the analyte or the stationary phase is minimal.
However, in HPLC, the mobile phase is a dynamic part of the system that influences both
analyte and stationary phase interactions.
In this method, enantiomeric separation is accomplished by the formation of a pair of
transient diastereomeric complexes between racemic analyte and the chiral mobile phase
additive. Chiral discrimination is due to differences in the stabilities of the diastereomeric
complexes, solvation in the mobile phase, and/or binding of the complexes to the solid
support. The three major approaches in the formation of diastereomeric complexes114 are:
(1) transition metal ion complexes (ligand exchange); (2) ion pairs; and (3) inclusion
complexes. A detailed discussion of each mode is given by Allenmark115 and Ahuja116.
Many racemic mixtures can be separated on conventional achiral LC columns by
using an appropriate chiral mobile phase additive.
Additives such as α, β, and γ-
cyclodextrins have been successful. Advantages of this technique117 are as follows: (1)
less expensive conventional LC columns can be used; (2) a wide variety of possible
additives are available; and (3) different selectivities from the chiral phases can be
obtained. However, the problems with this technique118 include: (1) many chiral additives
are costly, and sometimes, have to be synthesized; (2) the mode of operation is complex,
and (3) incovenient for preparative applications because the chiral additive must be
removed from the enantiomeric solutes.
2.1.4 Direct Chromatographic methods (Chiral Stationary Phases)
Enantiomeric separation by using chiral HPLC stationary phases (CSPs) is based on
the formation of transient diastereomeric analyte-CSP complexes between the enantiomers
and the chiral molecule that is an integral part of the stationary phase.
12
At present, there are over a hundred CSPs for HPLC that are commercially available.
According to Wainer119, there are five major classes of HPLC-CSPs based on the type of
analyte-CSP complexes formed.
The Type 1 or “Pirkle” phase forms analyte-CSP
complexes by attractive-repulsive interactions, mainly by π electron donor-acceptor
mechanisms.
The Type 2, exemplified by derivatized cellulose, involves attractive
interactions followed by inclusion into chiral cavities.
The Type 3 CSPs,
such as
cyclodextrin and crown ethers, form inclusion complexes. In the Type 4 CSP, the analyte
is a part of a diastereomeric metal complex (chiral ligand-exchange chromatography). The
Type 5 CSP is a protein, e.g., bovine serum albumin, and the analyte-CSP complexes are
based on the combination of hydrophobic and polar interactions. Table I shows some
examples of commercially available CSPs and the mode of chromatographic analysis that
can be carried out.
Table I. Classification of Chiral Stationary Phases (CSPs)
Type
Description
Examples
Mode (modifiers)
1
Pirkle-type (π-donor or
DNB-phenylglycine,
Normal phase (polar)
π-acceptors)
DNB-leucine,
naphthylalanine
2
Attractive interactions
Chiralcel OA, OB, OD,
followed by inclusion
OF, OJ
Normal phase (polar)
(derivatized cellulose)
3
Inclusion (cyclodextrins,
Cyclobond I, II, III;
Reversed phase (aqueous
polyacrylates,
Chiralpak OP, OT;
acetonitrile or methanol)
polyacrylamides, crown
Chiralcel CR
ethers)
4
Ligand exchange
Proline, hydroxyproline
Reversed phase (aqueous
buffers)
5
Proteins
Albumin, glycoprotein
Reversed phase (aqueous
buffers)
13
Type 1 CSP: Pirkle Type and Related CSPs
Pirkle’s group120 at the University of Illinois is generally acknowledged as the
inventor of Type 1 CSP, although various CSPs of similar type have been developed by
Japanese121-122 and European123-126 laboratories.
They have advanced this area by
developing amino acid derivatives containing π-acidic or π-basic aromatic group that are
ionically or covalently linked to aminopropyl silica.
Furthermore, Pirkle has done
extensive mechanistic studies utilizing computer modelling of diastereomeric complex
interaction and various homologous series studies to elucidate the chiral recognition
mechanism of the CSP.
Enantiomeric separation is achieved by the following interactions: (1) π-π interactions
between the π-donor and π-acceptor aromatic rings of racemic analyte and CSP, and vice
versa; (2) hydrogen bonding involving secondary amines and carbonyl groups on the CSP
with acidic proton, hydroxyl, and amino groups on the analyte; (3) dipolar interactions as
in dipole stacking; and (4) steric interactions arising from the bulky nonpolar groups
attached near the chiral center of the CSP that provide conformational control.
Figure 3
shows a Pirkle CSP of N-(3,5-dinitrobenzoyl)phenylglycine ionically or covalently bonded
to a silica support.
Type 1 CSPs can separate a wide spectrum of racemic compounds, such as alkyl and
aryl carbinols, aryl substituted hydantoins, lactams, succinamides, phthalides, sulfoxides,
sulfides, amides and imides. For racemic analytes having polar functional groups, e.g.,
amino, carboxyl, and hydroxyl, they have to be made less polar by derivatization because
underivatized they interact too strongly with the CSP. Derivatization involves conversion
of the amino group to amide, urea, or carbamate functionality; carboxyl conversion to
ester, anilide, or amide; and hydroxyl conversion to ester and carbamate126.
Recently,
Pirkle127 has developed a hybrid column with both π-acid (3,5-dinitrobenzoyl) and π-base
(naphthyl) groups.
It can resolve a variety of compounds, including underivatized
carboxylic acids.
The mobile phases used for Type 1 CSPs usually contains a nonpolar organic solvent
with various amounts of polar modifier, e.g., hexane and 2-propanol.
14
(a)
NO2
OH
o Si
+
-
NH3
O2C
OH
H H
N
C
C
Ph
NO2
O
(b)
NO2
O
OH
o Si
N
H
OH
H H
N
C
C
Ph
NO2
O
Figure 3. Type 1 CSP: Pirkle N-(3,5-dinitrobenzoyl)phenylglycine (a) ionically or
(b)
covalently bonded to a aminopropylsilica.
Type 2 CSPs: Derivatized Cellulose and Related CSPs
Cellulose is a crystalline polymer composed of a linear poly-β-D-1,4-glucoside.
Energy calculations support a helical structure in which each glucose residue lies near the
helical axis but is twisted relative to the previous one128. At the supramolecular level,
cellulose contains ribbonlike polymer chains that are held together by intramolecular and
intermolecular hydrogen bonds that results in the formation of sheets; the sheets then stack
15
vertically and are staggered 1/2 a glucose unit (Fig. 4). Thus, chiral cavities are created
between sheets, further imparting chiral character129. Unmodified cellulose was reported
to resolve amino acids and derivatives130-131, diaminocarboxylic acids132, synthetic
alkaloids133, and cathecins134.
Although crystalline cellulose shows chiral recognition, it does not make practical
CSPs135. Poor resolution and broad peaks are obtained due to slow mass transfer and
slow diffusion through the polymer network.
The highly polar hydroxyl groups
Figure 4. Structural diagram of unmodified cellulose showing the β-(1,4) linkage of Dglucose residues and intramolecular and intermolecular hydrogen bonding resulting in
fomation of sheets. (From ref. 129)
16
characteristic of cellulose often lead to nonstereoselective binding between the analyte and
cellulose.
In addition, cellulose is unable to withstand normal HPLC pressures.
Derivatization of cellulose brings about practically useful CSPs with high chiral
recognition mechanisms that can separate a wide range of racemic compounds. Okamoto
et al. developed the first commercially successful series of cellulose triacetates initially
coated136 and later bonded137 to a silica gel support. The other Type 2 CSPs that are
commercially available from the Daicel Chemical Industries have been developed by
Okamoto138 and co-workers. A list of commercially available Type 2 CSPs is given in
Table II.
The derivatized cellulose CSPs have been studied extensively140-143. However, it was
Wainer et al.144-145 in 1986 who first concluded that the chiral recognition mechanism,
which is generally accepted for most of the Type 2 CSPs, involves attractive interactions
followed by the steric fit of the molecules into the chiral surface.
That is, the analyte-
CSP complexes are formed first by attractive interactions and chiral discrimination then
takes place due to the steric fit of racemic analyte into the chiral cavity.
The attractive
interactions include hydrogen bonding, dipole-dipole, and π-π interactions.
Some derivatized amyloses146 (Table II) are also classified as Type 2 CSPs. They
exhibit similar behavior in discriminating racemic analytes. Amylose is a natural polymer
possessing the same constituents as cellulose but differing in the glycosidic linkage, in this
case a helical α-D-1,4-glucoside. The chiral recognition mechanism for the amylose
CSPs, as concluded by Wainer et al.147, involves attractive interactions and steric fit of
analytes into chiral cavities, but the latter is no longer critical for chiral discrimination.
The chiral recognition mechanism is said to be conformationally driven. If the derivative
group of an amylose CSP is the same as that of cellulose, the expected enantioseparating
ability of two CSPs should be complementary148.
Although amylose and cellulose have
the same D-glucose constituents but they have different higher order structures due to
difference in the glucose linkage. Fig. 2 shows the structure of the Type 2 CSPs used in
this research, Chiralpak AD Chiralcel OD. Both CSPs have the same derivatized glucose
constituents.
17
Table II. Commercially Available Type 2 Chiral Columns (After Okamoto, ref 139. )
Polysaccharide Derivative
Trade Name
Distributor
Cellulose triacetate (coated on silica gel)
Chiralcel OA
Daicel
Cellulose tribenzoate (coated on silica gel)
Chiralcel OB
Daicel
Cellulose trisphenylcarbamate (coated on
silica gel)
Chiralcel OC
Daicel
Chiralcel OD
Chiralcel OD-R
Daicel
Cellulose tris(4-chlorophenylcarbamate)
(coated on silica gel)
Chiralcel OF
Daicel
Cellulose tris(4-methylphenylcarbamate)
(coated on silica gel)
Chiralcel OG
Daicel
Cellulose tris(4-metylbenzoate) (coated on
silica gel)
Chiralcel OJ
Daicel
Cellulose tricinnamite (coated on silica
gel)
Chiralcel OK
Daicel
Amylose tris(3,5-dimethylphenylcarbamate (coated on silica gel)
Chiralpak AD
Daicel
Amylose tris [(S)- phenylethylcarbamate)
(coated on silica gel)
Chiralpak AS
Daicel
Cellulose tris(3,5-dimethylphenylcarbamate (coated on silica gel)
18
Type 2 CSPs of derivatized cellulose and amylose developed by Daicel Chemical
Industries, Japan, are reported to separate a wide range of racemic compounds. These
include axially and planar asymmetric compounds, metal containing compounds,
compounds possessing chiral sulfur or phosphorus atoms, hydrocarbons, halides, cyano
compounds, ketones, amines and their derivatives, acids and their derivatives, alcohols and
their derivatives, ethers and oligomers149.
Most of the commercially available derivatized
cellulose and amylose CSPs are adsorbed onto a macroporous silica gel support150. It is
believed that the higher order structure of a Type 2 CSP greatly influences its
enantioseparating ability.
Type 3 CSPs: Inclusion
Chiral separations based on inclusion are achieved through a mechanism by which the
guest molecule is included into the cavity of a host molecule. The exterior of the host
molecule generally possesses functional groups that act as steric barriers or interact with
with the guest molecule in a fashion that induces enantioselectivity.
The most often used CSP of this type is cyclodextrin bound to a silica support.
Cyclodextrins151 are cyclic oligomers with 6 (α), 7 (β), or 8 (γ) D-glucose units through
α-1,4-linkage that adopt a tapered cylindrical or toroidal shape (Fig. 5).
The toroid has
a maximum diameter ranging from 5.7 Å (α-cyclodextrin) to 9.5 Å (γ-cyclodextrin) with a
depth of about 7 Å. The interior of the toroid is hydroxyl free and hydrophobic which
favors enantioseparation of partially nonpolar compounds via selective inclusion. Since
inclusion is critical for chiral discrimination, α-cyclodextrin is suitable for small racemic
molecules and γ-cyclodextrin CSP is better for big molecules. Since cyclodextrin as a
whole is hydrophilic, aqeuous mobile phases are used. Derivatized cyclodextrins152 have
also been developed to vary the enantioseparating ability.
19
(a)
(b)
Figure 5. Type 2 CSP: (a) Structural diagram of β-cycldextrin and two of the glucose
residues that illustrate the details of the α-(1,4) glycosidic linkage and (b) the schematic
diagram of cyclodextrin bonded to a silica gel support forming an inclusion complex with
a chiral molecule. (From ref. 151)
.
20
Armstrong153-154 has studied extensively the cyclodextrins and popularized them as
both CSPs and mobile phase additives. Cyclodextrins possess excellent selectivities for a
wide range of racemic compounds: amines, alcohols, carboxylic acids,epoxides, and
others.
At present, cyclodextrin is most often used for chiral separation as a chiral
stationary phase in both GC and LC, and as mobile phase additives in LC, GC, and CE.
Crown ethers are classified as Type 3 CSPs. Unlike cyclodextrin, the host-guest
interaction within the chiral cavity is hydrophilic in nature. Crown ethers are heteroatomic
macrocycles with repeating units of (-X-C2H4-) where X, the heteroatom,is commonly
oxygen but may also be a sulfur or nitrogen atom. Crown ethers155, especially 18-crown6 ethers, can complex inorganic cations and alkylammonium compounds. This inclusion
interaction is based mainly on hydrogen bonding between the hydrogens of the ammonium
group and the heteroatom of the crown ether. Additional electrostatic interaction occurs
between the nitrogen and the crown ether’s oxygen lone pair electrons. Crown ether
CSPs are made up of crown ethers covalently bonded to a silica gel or polystyrene matrix.
It was Cram and co-workers156 who first demonstrated the enantioseparating ability of
crown ether for a number of amino esters and amino acids. Shinbo et al.157 found that
chiral separations can also be performed on a reversed phase column that was dynamically
coated with a crown ether. This crown ether has been utilized in a commercially available
column, Crownpak CR (Daicel Chemical Industries, Japan) and has been used to separate
a large number of amines, amino alcohols, amino acids, and amino esters.
Other Type 3 CSPs are made up polyacrylates and polyacrylamides as the chiral
selectors.
Type 4 CSPs: Ligand Exchange
As early as 1968, the ligand exchange technique first offered a complete and reliable
separation of enantiomers of amino acids and some chiral compounds that form complexes
with metal cations.
Ligand exchange chromatography (LEC)158 involves the formation of a reversible
coordination complexes between a bidentate analyte, a divalent metal ion, and a chiral
21
ligand immobilized in the stationary phase. The metal ion is a transition metal, usually
Cu+2, and the chiral ligand is an amino acid, e.g., proline. The stability of such complexes
is highly dependent on the transition metal used: Cu+2 complexes are generally the most
stable and are preferred in LEC applications. Factors affecting selectivity and efficiency of
the enantioseparations include the pH, ionic strength, and column temperature. Ligandexchange chromatography was first reported by Davankov’s group159-160 in Moscow in
1971. They used immobilized L-proline residues on styrene-divinyl benzene resins in the
presence of Cu+2 to separate racemic amino acids.
Gubitz et al.161 in 1983 improved efficiencies of the original Type 4 CSP by using a
silica gel, instead of the styrene-divinyl benzene packing material. Karger et al.162-163 in
1983 further improved the efficiency and enantioselectivity of such packing by
incorporating an inert n-butyl or n-decyl spacer chains between the ligand groups
interacting with the metal atom and the silica support.
Ligand exchange chromatography, however, is limited only to racemic analytes that
act as a bidentate ligand for Cu+2 or other comparable ions, e.g. amino acids. Alternatively
derivatization is needed for the enantiomers to be separated.
Type 5 CSPs: Proteins
Proteins are polymers composed of chiral units (L-amino acids) and are known to
bind small molecules. A protein can be immobilized on solid support by a variety of
bonding chemistries; the choice of bonding chemistry affects the selectivity of the CSP
obtained.
In 1973, Stewart and Doherty164 first reported the use of agarose-bonded
bovine serum albumin (BSA) for chiral resolution of racemate amino acids. This was
followed by the devolopment of other protein CSPs. Allenmark et al.165 in 1982 reported
agarose-bonded BSA, and Hermansson166 in 1983, a silica bonded α1-acid glycoprotein.
Other protein CSPs are ovomucoid167, human serum albumin168, avidin169, and
cellobiohydrolase170.
The retention mechanism for Type 5 CSPs is most likely based on hydrogen bonding,
ionic interactions, or other interactions171. The mobile phases are generally aqueous
22
buffers that are used in limited pH ranges with organic modifiers.
Retention and
stereoselectivity can be manipulated by varying the mobile phase composition, pH, and
column temperature. Although the capacity of protein column is limited, they offer a
broad range of applicability. The Type 5 CSPs are used to separate neutral, cationic, and
anionic racemic species.172
2.2 Temperature Dependence of Chiral Separation
Variation of column temperature has an effect on retention and enantioselectivity of
chiral analytes on a CSP. For example, an increase in temperature generally decreases the
retention of analytes due, on a molecular level, to their lower affinity with the stationary
phase and therefore their faster migration through the chromatographic column.
However, the enantioselectivity may either increase or decrease depending on the type of
interaction mechanism.
Thus, a change in temperature can be used to optimize
enantioseparation. From the retention behavior and measurements of thermodynamic
parameters - enthalpy, entropy, and Gibbs free energy of association between the
enantiomers and CSP, and enthalpy-entropy compensation analysis173 some information on
the mechanism for enantioseparation can be obtained.
The temperature studies in Chiral HPLC utilized the methods developed for reversed
phase HPLC and are sufficiently noteworthy to be reviewed here.
In 1978, Melander, Campbell and Horvath174 used the enthalpy-entropy compensation
method to conclude that the same retention mechanism was operating for the separation of
a homologous series of analytes on the different hydrocarbonaceous (reversed phase)
stationary phases at varying mobile phase compositions and temperature conditions.
These authors used the retention data collected at the different chromatographic
conditions and plotted the natural logarithm of capacity factors (another expression for
retention) against the reciprocal of absolute temperature (van’t Hoff plots). From the
slopes of van’t Hoff plots, the enthalpy changes for the partitioning of analytes between
stationary and mobile phases were measured.
A compensation plot (natural logarithm
of capacity factors against the measured enthalpies) was then constructed and it gave a
23
linear plot suggesting that a single, partitioning mechanism existed for the interaction
between analyte and the stationary phase.
Mathematically, the relationship for enthalpy-entropy compensation is given by
∆ H°° = β ∆ S°° + ∆ G°°β
(2.2a)
where ∆ G°°β is the Gibbs free energy change at the compensation temperature β and ∆ H°°
and ∆ S°° are the standard enthalpy and entropy changes, respectively. From the above
equation, the Gibbs free energy change is independent of temperature, since changes in
enthalpy are compensated by entropy. A detailed discussion of the theory of enthalpyentropy compensation method is given in Chapter III.
In 1986, Hancock et al.175 utilized column temperature data for the studies of protein
and peptide separations by reversed phase HPLC. The authors noted that, in contrast to
many biochemical systems in which an increase in temperature results in increased
interaction, in the separation of peptides an increase in temperature decreased the
retention. Furthermore, the separation of insulin proved to be entropy controlled, and the
separation of a lipid-binding peptide was enthalpy driven.
These conclusions were inferred by means of van’t Hoff plots176 obtained from
ln k = -∆
∆ H°° / RT + ∆ S°° / R + ln ϕ
(2.2b)
where k, is the retention factor (another expression for retention); ϕ is the volume ratio of
the stationary and mobile phases; R is the gas constant; T is the absolute temperature, and
∆ H°° and ∆ S°° are the enthalpy and entropy changes, respectively, when the analyte
interacts with the stationary phase. If the measured value of ∆ H°° is greater than ∆ S°°, the
separation is said to be enthalpy controlled. For enthalpy driven separations, lowering the
temperature enhances separation. The reverse is true for an entropy driven separation that
is characterized by a larger ∆ S°° value than ∆ H°°. A detailed discussion of this theory is
given in Chapter III.
Cole and Dorsey177 in 1992 utilized van’t Hoff plots and the measured enthalpy and
entropy parameters to study the retention mechanism in reversed phase HPLC. In their
study, the entropy contribution increased proportionally to the enthalpy contributions
when the bonding density of the sorbent surface was increased. The authors concluded
24
that partitioning, rather than adsorption, must be the “relevant model of retention” for the
analytes on a reversed phase stationary phase.
Although the above studies are for the separation of achiral compounds on achiral
stationary phases, they are discussed because of the use of temperature and vant Hoff
plots to help understand the retention mechanism involved. Similar experiments have been
used in this research to investigate the chiral recognition mechanisms for the CSPs used in
this research.
One of the first studies to utilize temperature for the optimization of chiral separation
was by Feitsma et al.178 in 1985. The enantioseparation of aromatic carboxylic acids on βcyclodextrin was improved by raising the temperature to 55°C. At 25°C enantiomeric
resolution of the acids was poor because the second eluting peak was broad and tailing.
There was an overall increase in resolution from 1.6 to 1.65 when the temperature was
55°C. Although there was no explanation for the increase in resolution, it appears that the
major effect may simply due to less adsorption at the higher temperature and reduced
tailing.
In 1986, Mazzo et al.179 optimized the chiral separation of an aromatic carboxylic
acid, MK-286, an anti-hypertensive drug at 0°C. The CSP was a Pirkle-type covalent
dinitrobenzoyl-L-phenylglycine using a mobile phase of methanol/aqueous buffer. The
resolution was improved from 0.75 to 1.25 when the temperature was lowered form 25°C
to 0°C. This is probably an enthalpy driven improvement, resulting from increased
selective interactions, like hydrogen bonding.
Pescher et al.180 in 1986 studied the temperature dependence, coupled with mobile
phase effects, of the enantioseparation of tertiary phosphine oxides on a Pirkle-type
dinitrobenzoylalanine CSP.
The mobile phase was hexane/chloroform/ethanol
(25/66.9/8.1). In this study the enantioselectivity was enhanced at lower temperature
which compensated for the loss in efficiency due to the increase in the mobile phase
viscosity (slower mass transfer). The enantioselectivity changed from 1.52 to 2.33 when
the temperature was decreased from 50°C to -15°C. The authors further noted that low
temperatures are useful for separating extremely thermolabile compounds.
25
An interesting temperature dependence study was reported by Weaner et al.181 in
1988. The enantioseparation was for fatty acid esters and amide epoxides on a Pirkle-type
CSPs using a hexane/2-propanol eluent system. Maximum resolution occured at different
temperatures for the ester and amide enantiomers, with the overall profile of resolution
versus temperature being an inverted parabola. For the ester, the maximum resolution was
1.6 between 0 and 5°C and for the amide, 1.7 at 38°C. Enantioselectivity of the ester
decreased rapidly above 0°C, but for the amide it reached a peak at 10°C and stayed
constant up to 40°C. These results were attributed to the weak analyte-CSP interaction
for the ester, in contrast to the strong (probably hydrogen bonding) amide-CSP
interaction.
From this study of Weaner et al.182 , it is important to point out that:
(a)
enantioseparation on a CSP is dependent on the solute structure which defines the type
and extent of interaction with the CSP; and (b) the nonlinearity dependence of
enantioselectivity with temperature, i.e. showing a maxima, can be attributed to the degree
of solute-CSP interaction.
Gilpin et al.183 in 1991, reported a maxima in plots of ln k versus the absolute
temperature (van’t Hoff plots). In the study, D- and L-tryptophan were enantioseparated
on a bovine serum albumin CSP, at varying temperatures and pH. The authors speculated
that the initial increase in ln k with temperature, which is “thermodynamically inconsistent
with a simple retention mechanism”, may be due to either gross changes in protein surface
orientation or induced conformational changes of the protein, which then provide an
increased number of binding sites.
It was concluded that the unusual curve and
nonlinearity of the van’t Hoff plots was due to a change of the retention mechanism.
Armstrong et al.184 in 1992 utilized only temperature dependence and calculated
thermodynamic parameters from van’t Hoff plots and enthalpy-entropy compensation
method to study the enantioselective retention mechanism of derivatized cyclodextrin GC
CSPs. A homologous series of TFA derivatized amines, TFA derivatized diols, and alkyl
esters of 2-bromobutyric acid were the probe molecules. Based on the results, the probe
molecules were divided into two groups: (a) Group I compounds that gave a nonlinear
compensation plot and a large enthalpy and entropy values, and (b) Group II compounds
26
that gave a linear compensation plot and lower enthalpy and entropy values. It was
concluded that the Group I compounds were enantioseparated by an inclusion mechanism
complex. On the other hand, the Group II compounds were resolved by external, multiple
association with the cycodextrin (top, side, and / or bottom surfaces). The authors further
noted that the enantioselective retention mechanism of some compounds may be
temperature dependent, e.g., more inclusion at lower temperatures.
The importance of Armstrong’s work for the research on chiral recognition
mechanism
lies in the possibile similarity of multiple retention mechanism between
cyclodextrin and the Type 2 CSPs, Chiralcel OD and Chiralpak AD.
It was noted
previously that the general recognition mechanism for the derivatized cellulose and
amylose CSPs involves both attractive interactions and inclusion or steric fit into chiral
cavities. If the attractive-repulsive interactions and inclusion mechanisms are competing,
temperature studies may determine which is dominant, e.g., van’t Hoff plots and
measurements of thermodynamic parameters.
There are additional reports on the use of column temperature to optimize HPLC
chiral separations. For the Type 2 CSPs, Witte et al.185 in 1992 reported the effect of
temperature on the enantioseparation of different aminotetralins on cellulose tris(3,5The resolution of aminotetralins with π-
dimethylphenyl-carbamate), Chiralcel OD.
electrons not directly bonded to the hydrogen bonding of the molecule was enhanced by
lowering the temperature.
This result was attributed to π-π interactions that are more
pronounced at lower temperatures. According to the authors, “the π-π interaction is such
a weak force that at higher temperatures the energy of the analyte and the CSP is so high
that the π-π interaction is easily disturbed”. It was also noted that the aminotetralins
which have either the primary or secondary amine group and possess a sufficiently short
N-alkyl substituent showed a decrease on resolution at lower temperatures, but all other
racemic amino aminotetralins showed an increase of resolution.
This result was
explained in terms of the hydrogen bonding interaction , which is a stronger force and less
influenced by temperature.
Another temperature study that is worth mentioning is that of Cabrera et al.186 in
1994.
Oxazepam and prominal were enantioseparated on a chemically bonded β-
27
cyclodextrin , LiChroCART Chiradex, using a reversed phase eluent. A decrease in
temperature caused an increase in the retention of both analytes, but the enantioselectivity
for the former was
increased and of the latter was decreased.
These results were
explained in terms of thermodynamic parameters that control the enantioseparation.
Oxazepam which favored the low temperature was said to be an enthalpy controlled
enantioseparation. Prominal, on the other hand, that favored a high temperature for the
enantioseparation was said to be entropy driven.
In addition, Smith, Taylor, and Wilkins187 in 1995 reported the chiral separations of a
series of related compounds within a series of potassium channel activator analogues at
different temperatures.
The CSP was cellulose tris(3,5-dimethylphenylcarbamate),
Chiralcel OD, and the mobile phase was a mixture of hexane and 2-propanol.
By
thermodynamic measurements, the analog that showed better enantioseparation at low
temperature was enthalpy controlled. This result was attributed to a hydrogen bonding
phenomenon that is important in the chiral discrimination. For the other analog that
favored the high temperature, its enantioseparation was said to be entropy controlled. The
authors concluded that the π-π interactions were important in the chiral recognition
process.
Lastly, Booth and Wainer188 in 1996 investigated the retention mechanism of the
amylose tris(3,5-dimethylphenylcarbamate) CSP, Chiralpak AD, for mexilitine and for a
series of structurally related compounds using the enthalpy-entropy compensation method
coupled with the quantitative structure enantioselective retention relationship (QSERR)
method. The enthalpy-entropy compensation analysis, which was used to determine a
multiple retention mechanism, suggested that there were two retention mechanisms for
the CSP. The authors further noted that these mechanisms were based on either the
presence or absence of secondary hydrogen bonding groups.
From the cited works, temperature dependence studies that include retention
behavior, measurements of thermodynamic parameters, enthalpy, entropy, and Gibbs free
energy of association between the enantiomers and CSP, and enthalpy-entropy
compensation analysis, provide important information for the enantioselective retention
28
mechanism. However, these studies do not give a complete description of the mechanism
for the separation of a chiral compound on a CSP.
2.3 Mobile Phase Studies (Polar Modifiers)
In liquid chromatography, the separation of a chiral compound on a CSP is thought of
an adsorption-desorption process189.
The mobile phase molecules compete with the
racemic analyte for the active binding sites of the CSP. After column equilibration, the
polar modifier molecules occupy the adsorption active sites of the CSP. As the analyte
“plug” enters the column, analyte molecules begin displacing mobile phase molecules from
interaction sites of the CSP. The relative affinities of the racemic analyte versus the
mobile phase species for the CSP active sites determine the retention, k, for a given
analyte, and the difference in affinities between the analyte enantiomers and the mobile
phase determine the enantioselectivity, α. A decrease in retention when different mobile
phase is used means that the analyte has lesser affinity for the CSP compared to the
different mobile phase, and the analyte molecules have a difficulty in displacing the
different mobile phase molecules.190
Mobile phase studies help elucidate the possible chiral recognition mechanism for a
CSP. Variation of the mobile phase composition enables one to help predict the dominant
interactions that influence retention, and the interactions that are critical to
enantioselectivity.
2.3.1 Alcoholic Mobile Phase Modifier
As shown in Table I, Type 1 and Type 2 CSPs generally utilize normal phase eluent
with a polar modifier, such as an alcohol. As noted in the previous discussion, the chiral
selectors of the Type 1 and Type 2 CSPs are polar and are capable of forming hydrogen
bonds. Alcohol as the polar modifier, by virtue of its functionality, is also capable of
forming hydrogen bonds.191 The alcohol molecules, therefore, compete with the analytes
for the active adsorption sites of the CSP.
29
In 1984, Zief et al.192 reported the enantioseparation of
2,2,2-trifluoro-1-(9-
anthryl)ethanol on covalent R-N-(3,5-dinitrobenzoyl)phenylglycine (a Pirkle-type CSP)
using hexane with either ethanol, 2-propanol, or tert-butanol as the polar modifier. The
polarity of the alcoholic mobile phase modifier was kept constant, thus, the only difference
lay in the steric structure of the alcohols. Of the three alcohols tested tert-butanol gave
the best enantioselectivity, α = 1.62 vs. α = 1.56 for 2-propanol and α = 1.33 for ethanol.
This increase in α was attributed mainly to an increase in the k of the primarily bound
enantiomer while the k of the first eluted enantiomer was essentially unchanged. The
authors concluded that an increase in the bulk of the alcohol decreased its absorption on
the CSP.
Pescher et al.193 in 1986 reported similar results in the resolution of enantiomeric
phosphine oxides on Pirkle-type dinitobenzoylalanine CSP. The mobile phases used were
a mixture of hexane and alcohol, and the polarity of the alcoholic mobile phase modifier
was kept constant. The enantioselectivity of the system increased with increasing bulk of
the alcohol.
Macaudiere et al.194 in 1986 reported the resolution of a series of enantiomeric
amides using a Pirkle-type CSP under SFC conditions.
The amides were
chromatographed using carbon dioxide modified with an alcohol. The results of the study
indicated that at constant k, the stereoselectivities are greater for alcohols with bulkier side
groups close to the hydroxyl moiety.
So far, all studies cited are about the influence of alcoholic mobile phase modifiers on
k and α on a Pirkle-type CSP. In all these studies, the enantioselectivity increases with an
increasing bulk of the modifier. In practical terms, this means that the bulkiest modifier
should be used to optimize chiral separations.
Studies of the effect of the alcoholic mobile phase modifiers for the Type 2 CSPs have
also been made. Mobile phase effects on k and α on the cellulose triacetate CSP were
investigated by Koller et al.195 in 1983. In their study, the eluent was changed from
methanol to ethanol-water (96:4) to 2-propanol.
Three enantiomeric solutes were
chromatographed. The k of each solute increased throughout the study while the effect on
α varied according to the structure of the solute. The authors concluded that the polarity
30
of the eluents may not be the key to their elution power. It was further noted that if the
analytes were included into chiral cavities of the cellulose triacetate CSP, the elution
power may be determined primarily by the size of the eluent. This suggests that the bulk
of the alcoholic mobile phase modifiers affects the steric environment of the chiral cavities.
It must be remembered that the higher order structures of the derivatized cellulose
influences the chiral separation process.
Wainer et al.196 in 1987 investigated the effect of the bulk of an alcoholic mobile
phase modifier on k and α of a series of amides and alcohols on the cellulose tribenzoate
CSP.
The mobile phases used consisted of hexane and alcohols.
At a constant
concentration of ethanol in hexane, k of each solute increased with decreasing polarity of
the alcoholic modifier while α varied according to the structure of solute. The results of
their studies indicated that as in the Pirkle-type CSP both the analyte and alcoholic mobile
phase modifier compete for chiral and achiral binding sites on the CSP. The bulk around
the hydroxyl moiety of the alcoholic mobile phase modifiers played a role in this
competition. In addition, the results also suggested that like the cellulose tribenzoate CSP,
the alcoholic mobile phase modifiers may bind to sites near or at the chiral cavities of the
CSP changing the steric environment at these cavities and, thus, the stereoselectivity of the
CSPs.
2.3.2 Mobile Phase Effects (Acidic Modifier)
The majority of the enantioseparations of chiral acidic and basic compounds on Type
2 CSPs require the addition of acidic mobile phase modifiers197-198. However, there have
not been any in depth studies on the influence of the acidic mobile phase modifiers on
retention and enantioselectivity in Type 2 CSPs. Okamoto et al.199 in 1989 reported the
successful
enantioseparation
of
2-arylpropionic
acids
on
tris(3,5-
dimethylphenylcarbamate)s of amylose and cellulose using 1% trifluoroacetic acid (TFA)
as the acidic mobile phase modifier. These authors mentioned that TFA helped in the
adsorption-desorption process. In 1996, Booth et al.200 studied the chiral discrimination
mechanism in amylose tris(3,5-dimethylphenylcarbamate), Chiralpak AD, for 2-
31
alkylarylpropionic acids and 1% TFA as the acidic modifier. Tang201 in 1996 attempted
to explain the role of TFA on the enantioseparation of ibuprofen and ketoprofen on tris(4methylbenzoate), Chiralcel OJ. According to her results, small amounts of TFA only
improved the enantiomeric resolution of both analytes because they were already partially
resolved without TFA. Tailing peaks without TFA and longer retention times were due to
the strong interactions of ibuprofen and ketoprofen with the accessible silanol groups of
the packing material. That is, TFA essentially deactivated the CSP resulting in less peak
tailing and therefore, better resolution. Furthermore, Tang explained that TFA displaced
the analytes acidic group on silanol sites of CSP and suppressed the ionization of the
weakly acidic silanol groups.
2.4 Chiral Recognition Mechanism Studies
Predicting the chiral recognition of a CSP for a particular racemic compound is a
difficult task because of the complexity of the process. The preferential interaction of a
CSP with one of the enantiomers, known as chiral recognition, reflects only those sets of
interactions that are important for enantioselectivity causing differential retention of
enantiomeric analytes.
2.4.1 Studies Involving Variation of Structural Features of a Common Skeleton
In this method a homologous or analogous series of chiral compounds are used as test
probes carefully designed to interact with the chiral stationary phase. Retention factors
and enantioselectivities are evaluated and then correlated with the structural features of
the probe molecules.
As long as the differences in retention are substantial, this method
can be used with confidence and has the advantage of requiring only very small amounts
of compounds.
In 1980, Pirkle et al.202 used the chiral derivatives of a 3,5 dinitrobenzoylamines,
alcohols and thiols as test probes for a fluroanthrylalcohol CSP covalently bonded to a
silica support. The results obtained confirmed the theory of π-π interactions between the
32
nitrobenzoyl and anthracene moieties (π-electron acceptor and donor, respectively), and
hydrogen bonding between the benzoyl carbonyl group and the hydroxyl proton on the
CSP.
Hesse and Hagel203, Blaschke204, and Francotte et al.205, to cite works specifically on
derivatized cellulose CSPs, used various racemic phenyl-substituted analytes to study the
chiral recognition mechanism of microcrystalline cellulose triacetyl (CTA-I). The authors
proposed that the mechanism involved the penetration of the aromatic portion of the
solute into the cavities that are formed between the D-glucose units of the CSP. The
observed stereoselectivity was due to differences in fit or inclusion of the enantiomers of
the analytes in the cavities of the CSP. The mechanism is similar to the inclusion (hostguest) mechanism proposed by Armstrong et al.206 for the cyclodextrin-based CSPs.
Wainer et al. in 1986207, studied the chiral recognition mechanism for another type of
derivatized cellulose CSP.
Chiral amides were used and the CSP was cellulose
tribenzoate. The results indicated that the formation of the diastereomeric analyte-CSP
complex is based on hydrogen bonding, π-π and amide dipole interactions.
These
interactions not only formed the complexes but also positioned the analyte on the CSP. In
addition, the magnitude of the enantioselectivity of the enantiomeric amides was
dependent upon the differences in the steric fit of the asymmetric portion of the analyte in
a chiral cavity of the CSP. It should be noted that the proposed mechanism here is
different from that of CTA-I, which is the formation of inclusion complex.
Since the higher structural order of the derivatized cellulose CSPs influences its
enantioseparating ability, Wainer and co-workers208 followed up their study by exploring
the influence of mobile phase on retention and enantioselectivity. This was to verify if
steric fit into chiral cavity was critical to chiral discrimination. Inclusion in this context
means that the formation of diastereomeric complexes involves inclusion of analytes in
chiral cavities followed by attractive interactions. It must be remembered that the higher
order structure of derivatized cellulosed CSPs affects its enantioseparating ability.
In 1987, Wainer et al.209 investigated again the chiral recognition of the same CSP
but for chiral aromatic alcohols. In this study, both the influence of temperature and
mobile phase were explored. They concluded that a similar mechanism existed, where the
33
first attractive interaction was hydrogen bonding between the alcohol moiety of the
analyte and the carbonyl oxygen of the ester moiety.
In 1992, Witte et al.210 studied the chiral recognition mechanism for the cellulose
tris(3,5-dimethyphenylcarbamate), Chiralcel OD using twenty one different aminotetralins.
In the study, the effects of temperature and mobile phase were also investigated. As
mentioned before, hydrogen bonding was the dominant interaction between CSP and
solute. Furthermore, according to the authors, the analyte-CSP “free” π-π interaction
greatly enhanced the separation.
Overall, the use of structurally related compounds can allow inferential conclusions to
be drawn about mechanisms of interactions.
The method is reliable as long as the
chromatographic data is reproducible and of high precision. In addition, if other studies
are performed such as temperature dependence and mobile phase effects, a better
understanding of the chiral recognition process by correlation of retention and
enantioselectivity using homologous or analogous series can be achieved. The information
obtained from the study of specific molecular interactions operating during a
chromatographic
process
may
be
exploited
to
gain
the
ability
to
predict
enantioseparations, as well as to further the development of highly efficient and even
customized CSPs.
2.4.2 Thermodynamic Studies
The basic concept was described earlier in Section 2.2. Thermodynamic approaches
for the study of chiral recognition mechanism also utilize homologous or analogous series
of compounds, and the retention behavior and the corresponding enantioselectivities at
different temperatures are evaluated.
The thermodynamic parameters, free energy,
enthalpy, and entropy of association between the enantiomers and CSP are then
calculated. Entropy and enthalpy data from the graphs of the logarithm of capacity factor
versus the reciprocal of absolute temperature (van’t Hoff plots) are utilized to indicate the
possible chiral recognition mechanism. However, the results of this study do not give a
complete description of the chiral recognition process. The study could best evaluate the
34
number of mechanisms for a CSP, e.g. single or multiple mechansms211, and whether the
enantioseparation is enthalpy or entropy controlled.
2.4.3 Quantitative Structure Enantioselective Retention Relationships (QSERR)
The QSERR approach to study the chiral recognition process also utilizes a
homologous or analogous series of chiral compounds carefully designed to interact with
the chiral stationary phase. The retention data are subjected to multiparameter regression
analysis or factor analysis against various non-empirical molecular descriptors, which
reflect the structural features of the analytes. Directed by the initial QSERR equations,
molecular modelling is then used to either expand or reject the hypotheses proposed via
the equations in order to predict retention and enantioselective separations based on the
structures of the solute and the CSP. Only a few studies have been made using this
approach because of its complexity and its requirement of a sophisticated computer
program.
This approach is generally termed as Quantitative Structure Retention
Relationship212 (QSRR) which was first developed in reversed phase HPLC.
213
chromatographers, Kaliszan
214
, Jandera
215
, and Jinno
Three
have been resposible for the
development of this approach.
Kaliszan et al.216 used this approach to study the retention and chiral recognition
mechanism for a series of benzodiazipines chromatographed on a human serum albumin
CSP. The QSERR analysis also provided information on the structure of the HSAbenzodiazepine binding site, which was consistent with the structure derived from X-ray
crystallographic studies.
In 1996, Booth and Wainer217 investigated the chiral recognition mechanism of the
amylose tris(3,5-dimethylphenylcarbamate) CSP, Chiralpak AD, for α-alkyl arylcarboxylic
acids using the QSERR method. Their study indicated that the chiral recognition process
involves initial hydrogen bonding interactions between the solute and CSP, insertion of a
solute into a cavity on the surface of the CSP, and the stabilization of the analyte-CSP
complex by formation of an additional hydrogen bond within the cavity. The chiral
recognition process was said to be “conformationally driven”.
35
Booth and Wainer218 also studied the chiral recognition mechanism of mexiletine an a
series of structurally related compounds on Chiralpak AD. By application of QSERR
coupled with thermodynamic approaches, two separate mechanisms were identified. As
mentioned before, the mechanisms are based on either the presence or absence of
secondary hydrogen-bonding groups.
The QSERR approach is valuable to chiral recognition mechanism studies because it
may describe exactly which part of the analyte and CSP is involved and to what extent this
is responsible for the separation. However, the disadvantages of this model include its
complexity, the need for specialized computer hardware and software, and the expertise to
use it. Lack of descriptors for the mobile phase molecules may also be a problem.219
2.4.4 Enthalpy-Entropy Compensation
A complementary approach with QSERR to investigate chromatographic retention
mechanism involves enthalpy-entropy mechanism220. Analogously to QSEERs, enthalpyentropy compensation manifests itself as a linear dependence of the overall free energy
change on the corresponding enthalpy change for intrinsically similar physico-chemical
phenomena.
Linearity in plots of logarithm of capacity factors, measured at an
appropriate reference temperature, and the corresponding enthalpies for the particular
chromatographic process
are indicative that the analytes display similar retention
mechanism in a given system. Like the QSERR, this method requires test probes of
similar structures.
The enthalpy-entropy compensation method, like the thermodynamic approach, is
used mainly to predict a single or multiple mechanism for a CSP. This does not give a
specific description of the chiral recognition process. The detailed methodology of this
approach is discussed in Chapter III.
36
2.4.5 Molecular Modelling
Molecular modelling studies are directed towards the understanding where and how
chiral discrimination takes place in a CSP. Most of the microscopic modelling for the
chiral recognition mechanism of a CSP utilizes theory which implements the following:
quantum mechanics, molecular mechanics, molecular dynamics, Monte Carlo simulations,
and molecular graphics221. Several molecular modelling studies have been made for all the
CSPs. An example is the molecular modelling studies done by Camilleri et al.222 for the
enantioseparation of oxiracetam and related molecules on cellulose triphenyl carbamate
CSP. In the study, calculations of minimum energy configurations were performed for the
first individual molecules, then various conformations for the docked pair. With the aid of
molecular modelling, the authors were able to identify the possible interactions leading to
stereoselection.
For the other molecular modelling studies, the reader is directed to the Lipkowitz224
review in 1995.
A shortcoming of the computer modelling approach is the lack of considerations for
the mobile phase effects. It must be noted that the mobile phase greatly influences in the
enantiomeric separation on a CSP. In addition, this is a complex method and requires a
sophisticated computer hardware and software, as well as the expertise of the analyst.
37